Nvidia launches multi-tasking chip for data centres



Jensen Huang, founder and CEO of NVIDIA Jensen Huang, founder and CEO of NVIDIA

Nvidia Corp (NASDAQ: NVDA), the California-based semiconductor firm, has unveiled the A100, a multitasking chip that can be split digitally to run several different programs on a single physical chip.

The chip is designed to help data centre owners increase computer power and competes with the capability of many Intel Corp chips. But the technology has been mostly limited to processor chips from Intel and similar chips, such as those from Advanced Micro Devices Inc. Nvidia said Thursday that its new A100 chip can be split into seven “instances”.

“Because it’s fungible, you don’t have to buy all these different types of servers. Utilisation will be higher,” said Nvidia chief executive, Jensen Huang.

On the US$200,000 price tag for a Nvidia DGX server built around the A100 chips, Huang added: “You’ve got 75 times the performance of a $5,000 server, and you don’t have to buy all the cables.”

Nvidia’s DGX server built around the A100 chips costs US$200,000. However, Nvidia said it hopes the A100 can replace the chips it sells for AI tasks, being used as a big single chip for training and split into smaller inference chips.


Newsletter

Time is precious, but news has no time. Sign up today to receive daily free updates in your email box from the Data Economy Newsroom.


The news came on the same day that Nvidia named CyrusOne (NASDAQ: CONE) a certified North American colocation partner for the NVIDIA DGX-Ready Data Centre program.

The scheme is intended to help global businesses and organisations access modern data centre facilities to build out their AI initiatives and infrastructure, and simplify deployment and allows CyrusOne to offer advanced data centre facilities in major markets for businesses that do not wish to build and deploy their own AI data centres.

“As AI systems continue to grow in complexity, size and power, a new generation of power-efficient computing hardware and underlying data centre cooling capabilities are required to handle this level of density and scale,” said John Gould, chief commercial officer at CyrusOne.

“As more critical workloads centralise in this large scale, high density compute, CyrusOne finds itself sitting at the epicentre of this data gravity with our nearly 50 data centres strategically located throughout the world. As we move into this new decade, data centres will continue to be an essential infrastructure to enable AI, machine learning, and help companies unleash the full potential of this technology.”

Read the latest from the Data Economy Newsroom: